Effective hyperparameter optimization using Nelder-Mead method in deep learning
Yoshihiko Ozaki, Yano Masaki, Masaki Onishi · IPSJ Transactions on Computer Vision and Applications · 2017
Abstract In deep learning, deep neural network (DNN) hyperparameters can severely affect network performance. Currently, such hyperparameters are frequently optimized by several methods, such as Bayesian optimization and the covariance matrix adaptation evolution strategy. However, it is difficult for non-experts to employ these methods. In this paper, we adapted the simpler coordinate-search and Nelder-Mead methods to optimize hyperparameters. Several hyperparameter optimization methods were compared by configuring DNNs for character recognition and age/gender classification. Numerical results demonstrated that the Nelder-Mead method outperforms the other methods and achieves state-of-the-art accuracy for age/gender classification.